Predicting the success of fragment screening by X-ray crystallography

Fragment screening using X-ray crystallography is a method that can provide direct three-dimensional readouts of the structures of protein-small molecule complexes for lead development and fragment-based drug discovery. With current technology, an amenable crystal form can be screened crystallographically against a library of 1000-2000 fragments in 1-2 weeks. We have performed over a dozen crystallographic screening campaigns using our own compound collection called Fragments of Life™ (FOL). While the majority of our fragment screening campaigns have generated multiple hits, some unexpectedly turned out to be nonproductive, either yielding no bound ligands, or only those thought to be inadequate for lead development. In this chapter, we have attempted to identify one or more parameters which could be used to predict whether a crystallized protein target would be a good candidate for fragment hit discovery. Here, we describe the parameters of crystals from 18 fragment screening campaigns, including six unsuccessful targets. From this analysis, we have concluded that there are no parameters that are absolutely predictive of fragment screening success. However, we do describe a parameter we have termed pocket factor which provides a statistically significant variance between nonproductive targets and productive targets shown to bind fragments. The pocket factor is calculated using a novel method of consensus scoring from three distinct pocket-finding algorithms, and the results may be used to prioritize targets for fragment screening campaigns based on an initial crystal structure.